Peris et al. BMC Genomics 2012, 13:407http://www.biomedcentral.com/1471-2164/13/407

RESEARCH ARTICLE Open Access

Comparative genomics among Saccharomycescerevisiae × Saccharomyces kudriavzevii naturalhybrid strains isolated from wine and beerreveals different originsDavid Peris1, Christian A Lopes2,3, Carmela Belloch2, Amparo Querol2 and Eladio Barrio1*

Abstract

Background: Interspecific hybrids between S. cerevisiae × S. kudriavzevii have frequently been detected in wine andbeer fermentations. Significant physiological differences among parental and hybrid strains under different stressconditions have been evidenced. In this study, we used comparative genome hybridization analysis to evaluate thegenome composition of different S. cerevisiae × S. kudriavzevii natural hybrids isolated from wine and beerfermentations to infer their evolutionary origins and to figure out the potential role of common S. kudriavzevii genefraction present in these hybrids.

Results: Comparative genomic hybridization (CGH) and ploidy analyses carried out in this study confirmed thepresence of individual and differential chromosomal composition patterns for most S. cerevisiae × S. kudriavzeviihybrids from beer and wine. All hybrids share a common set of depleted S. cerevisiae genes, which also aredepleted or absent in the wine strains studied so far, and the presence a common set of S. kudriavzevii genes,which may be associated with their capability to grow at low temperatures. Finally, a maximum parsimony analysisof chromosomal rearrangement events, occurred in the hybrid genomes, indicated the presence of two maingroups of wine hybrids and different divergent lineages of brewing strains.

Conclusion: Our data suggest that wine and beer S. cerevisiae × S. kudriavzevii hybrids have been originated bydifferent rare-mating events involving a diploid wine S. cerevisiae and a haploid or diploid European S. kudriavzeviistrains. Hybrids maintain several S. kudriavzevii genes involved in cold adaptation as well as those related toS. kudriavzevii mitochondrial functions.

BackgroundThe development of molecular methods of yeastcharacterization has demonstrated that some wine andbrewing Saccharomyces strains possess complex genomescomposed by genetic elements from two or more species[1-7]. These strains are widely known as interspecifichybrids.The best characterized industrial interspecific hybrid

is the lager yeast S. pastorianus, originated fromhybridization between S. cerevisiae and a S. bayanus-related yeast, which recently has been suggested to belong

* Correspondence: eladio.barrio@uv.es1‘Cavanilles’ Institute of Biodiversity and Evolutionary Biology, University ofValencia, Parc Científic, P.O. Box 22085, E-46071, Valencia, SpainFull list of author information is available at the end of the article

© 2012 Peris et al.; licensee BioMed Central LtCommons Attribution License (http://creativecreproduction in any medium, provided the or

to the new species S. eubayanus [8]. The hybridization be-tween S. cerevisiae and the cryotolerant S. eubayanus havebeen suggested as the result of selective pressures derivedfrom brewing at low temperatures [8].Other kind of natural Saccharomyces hybrids are those

originated from hybridization between S. cerevisiae andS. kudriavzevii. These hybrids have mainly been isolatedfrom wine and brewing environments [1-3].The role of the S. kudriavzevii genome in these

hybrids is unclear, since the known strains of this specieshave been found in decaying leaves from Japan and oaktrees from Portugal and Spain [9,10], but not in fermen-tative industrial environments yet. The physiologicalevaluation of some of these S. kudriavzevii isolatesshowed that this species is characterized by a higher

d. This is an Open Access article distributed under the terms of the Creativeommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andiginal work is properly cited.

Peris et al. BMC Genomics 2012, 13:407 Page 2 of 14http://www.biomedcentral.com/1471-2164/13/407

cryotolerance than S. cerevisiae, but a lower ethanol tol-erance [11,12].Albeit differences between S. cerevisiae × S. eubayanus

and S. cerevisiae × S. kudriavzevii hybrids, the role ofthe S. eubayanus or S. kudriavzevii genomes in the hy-brid seems to be similar, that is, the maintenance ofgood fermentative performance at low temperatures.The characterization of a particular group of Swiss

wine hybrids by PCR-RFLP, DNA arrays, ploidy analysisand gene dose determination by quantitative real-timePCR, evidenced the existence of a single commonhybridization event to explain the origin of these hybridsfollowed by extensive chromosomal rearrangements in-cluding chromosome losses and the generation of chi-merical chromosomes [13].In this work, genome composition by array-CGH of

a more diverse set of wine and brewing S. cerevisiae ×S. kudriavzevii natural hybrids from diverse origins wasevaluated to decipher their origins and evolution. Theexamination of gene losses and gains as well as themaintenance of specific metabolic pathways from theS. cerevisiae or S. kudriavzevii parental genomes wasalso analyzed with the aim of elucidating the role ofeach parental genome in the fermentative performanceof the hybrid strains.

MethodsYeast strains and culture mediaThe natural yeast hybrids S. cerevisiae × S. kudriavzeviiused in this study have been isolated from wine andbrewing fermentations in different locations (Table 1).The haploid strain S. cerevisiae S288c was used as con-trol for microarray DNA hybridizations. Yeast strainswere grown at 28°C in GPY medium (2% glucose, 0.5%peptone, 0.5% yeast extract).

Table 1 List of hybrid strains used in this study

Strain name Isolation source

HA1841 wine, Perchtoldsdorf, Austria

HA1842 wine, Perchtoldsdorf, Austria

PB7 wine Pietro Picudo, León, Spain

Assmanhausen (AMH) wine, Geisenheim, Germany

Anchor VIN7 commercial strain, Anchor, South Africa

SOY3 wine, Daruvar, Croacia

CECT1388 ale beer, United Kingdom

CECT1990 beer, Göttinger Brauhaus AG, Germany

CECT11002 beer Chimay Trappist, Belgium

CECT11003 beer Orval Trappist, Belgium

CECT11004 beer, Westmalle Trappist, Belgium

CECT11011 brewery, New Zealand

Ploidy estimations by flow cytometryPloidy estimates are very important to interpret aCGHdata from hybrids because hybridization signals are com-monly normalized with respect to those of the referencehaploid strain S288c.The DNA content of both hybrid and control strains

was assessed by flow cytometry by two different proce-dures. The first ploidy estimates were obtained in aFACScan cytometer (Becton Dickinson Inmunocytome-try Systems, California, United States) by using thepropidium iodide dye method described in Bellochet al. [13]. Due to discrepancies with the aCGH ana-lysis, new estimates were later obtained in a BeckmanCoulter FC 500 (Beckman Coulter Inc., California, USA)by using the SYTOX Green dye method described inHaase and Reed [14]. In both cases, ploidy levels werescored on the basis of the fluorescence intensity comparedwith the haploid (S288c) and diploid (FY1679) referenceS. cerevisiae strains. Ploidy reported for each strain isthe result of three independent measures. Results weretested by one way ANOVA and Tukey’s HSD tests.

DNA labeling and microarray competitive genomehybridizationTotal DNA, extracted as described in Querol et al. [15],was resuspended in 50 μl of de-ionized water anddigested with endonuclease Hinf I (Roche AppliedScience, Germany), according to the manufacturer’sinstructions, to fragments of an average length of 0.25 to8 kbp. Each sample was purified using High Pure PCRProduct Purification Kit (Roche Applied Science,Germany) and 2 μg was labelled using BioPrime ArrayCGH Genomic Labelling System (Invitrogen, California,USA). Unincorporated label was removed usingMinElute PCR Purification Kit (Qiagen, Germany). Equalamounts of labelled DNA from the corresponding hy-brid strains and the control S288c strain were used asprobes for microarray hybridization.Array competitive genomic hybridization (aCGH) was

performed using a double-spotted array containing 6,240ORFs of S. cerevisiae plus control spots totaling 6.4 K(Microarray Centre, University Health Network,Toronto, Canada). New microarrays were pre-treated forone hour at 65°C with pre-hybridization solution (7.5 ml20× SSC, 0.5 ml 10% SDS, 0.5 ml 10 mg/ml bovineserum albumin in 50 ml final volume). Pre-hybridizationsolution was washed during 15 s in mili-Q H2O, 2 s in2-propanol, 2 s in milli-Q H2O and dried by centrifuga-tion at 1200 rpm, 10 min. Microarrays were treated withhybridization solution (15 μl SSC, 0.6 μl 10% SDS, 6 μl1 mg/ml salmon DNA and DNA labelled in 60 μl finalvolume) at 95°C for 1 min and at room temperature for5 min before DNA hybridization. Hybridization was per-formed for 18 h in chamber at 65°C, thus allowing

Peris et al. BMC Genomics 2012, 13:407 Page 3 of 14http://www.biomedcentral.com/1471-2164/13/407

hybridization of the S. cerevisiae part of the hybrid gen-ome. A negative control of microarray hybridization wasdone by using DNA from S. kudriavzevii IFO 1802 strainvs. S288c. After hybridization microarrays were washedat 65°C for 5 min in 2× SSC, 0.1% SDS, at roomtemperature in 0.1× SSC− 0.1% SDS for 10 min and sixtimes in 0.1× SSC 1 min and dried by centrifugation at1200 rpm, 10 min.Experiments were carried out in duplicates and Cy5-

dCTP and Cy3-dCTP dye-swap assays were performedto reduce dye-specific bias. The aCGH was performedfor all hybrid strains except for W27, W46, SPG16-91and SPG441 previously analyzed by Belloch et al. [13].

Microarray scanning and data normalizationMicroarray scanning was done by using a GenePixPersonal 4100A scanner (Axon Instruments/MolecularDevices Corp., California, USA). Microarray images andraw data were produced with the GenePix Pro 6.1 soft-ware (Axon Instruments/Molecular Devices Corp., Cali-fornia, USA) and background was subtracted by applyingthe local feature background median option. M-A plots(M=Log2 ratios; A = log2 of the product of the inten-sities) were represented to evaluate if ratio data were in-tensity-dependent. The normalization process andfiltering were done using Acuity 4.0 (Axon Instruments/Molecular Devices Corp., California, USA). Rawhybridization signals from hybrids were normalized withrespect to those of the reference haploid strain S228cby using the ratio-based option, in which averagehybridization ratios are adjusted to 1 (and hence, thecorresponding log2 values to 0).Normalized data were filtered by regression correla-

tions 635/532 > 0.6, signal intensity in both channelsmore than 350 units, and signal to noise (SNR) > 2.5.Features with artifacts or flagged as bad were removedfrom the analysis. Replicates were averaged after filter-ing. It is worth to remark that strong normalization fac-tors were applied to the negative control signal in eachchannel (2 to the red and 0.46 to the green one). Rawdata and normalized microarray data are available inArrayExpress [16], under the ref. E-MEXP-3114.

Chromosome structure and recombination sites in thechimerical chromosomesThe log2 of normalized Cy5/Cy3 signal ratio obtained foreach ORF was represented with respect to its correspond-ing chromosomal location using the completely sequencedreference S. cerevisiae strain S288c. These plots, called car-yoscopes, were generated using ChARM v.1.1 [17]. Highlystringent hybridization conditions (65°C) were used toavoid the cross hybridization of S. kudriavzevii DNApresent in the hybrids. The caryoscope of the negativecontrol experiment showed that most S. kudriavzevii

genes did not hybridize under these conditions and in thecase of cross hybridization (red signal) this was due to thevery strong normalization factors applied in these control,which increased the red signal and reduced the green oneby factors not applied in the case of the experiments per-formed with DNA from hybrids (see Additional file 1:Figure S1). Accordingly, differences in the log2 ratio valuesobserved in the caryoscopes revealed variations in therelative copy number of S. cerevisiae genes present in thehybrid strains.The identification of over- and underrepresented

regions was confirmed due to the normalization proced-ure, the hybridization ratios derived from aCGH analysisshow the relative proportions of each gene with respect tothe average number of copies in the hybrid, allowing theidentification of over- and underrepresented regions in thehybrid genome by a one-way ANOVA test to determinethe different levels of hybridization observed in the aCGHanalysis. The approximate locations of the recombinationpoints in the mosaic chromosomes were determined fromthe up and down jump locations in the ORFs mapping bymicroarray analysis of the hybrid yeast genomes.Finally, by considering the collinearity of S. kudriavze-

vii and S. cerevisiae genomes [18], the S. kudriavzeviigene content in the hybrid genomes can be deducedfrom the presence/absence of the chromosome regionscoming from each parental species, obtained in a previ-ous PCR-RFLP analysis of these hybrids [19].

Gene Ontology (GO) analysis of S. kudriavzevii genesGenMAPP v2.1 software [20] was used to perform geneontology analysis of the S. kudriavzevii fraction in thehybrid genomes. Four different GO analyses were car-ried out using S. kudriavzevii genes present in all hybridstrains, including those previously characterized [13],these analyses corresponded to: i) the complete set ofwine and brewing hybrids, except strain AMH, showingthe lowest S. kudriavzevii gene content, ii) only winehybrids, except AMH, iii) only brewing hybrids and iv)only AMH. In all cases, statistically significant GO termenrichments were shown by computing a p-value usingthe hypergeometric distribution (the background set ofgenes was 6241, the number of ORFs measured inmicroarray experiments). GO terms showing significantvalues (z-score >2 and p-value <0.05) were sortedaccording to their corresponding GO category.

Maximum parsimony treeA list of minimal number of chromosomal rearrange-ments, chromosomal losses and restriction site changeswere used to reconstruct the maximum parsimony tree.Data obtained from a previous study [13] were againincluded in this analysis. A binary matrix was constructedto codify each particular event (Additional file 2: Table S1).

Peris et al. BMC Genomics 2012, 13:407 Page 4 of 14http://www.biomedcentral.com/1471-2164/13/407

Parsimony trees were constructed by PHYLIP 3.66 packageusing the Mix program [21], taking chromosomal rearran-gements and gain/losses as irreversible events (Camin-Sokalmodel) and the RFLP changes as reversible events (Wagnermodel). The consensus tree was obtained with Consenseprogram using the Majority rule.

Table 2 DNA contents of natural hybrids, estimated byflow cytometry using the SYTOX green method withrespect to the reference haploid and diploid S. cerevisiaestrains, S288c and FY 1679, respectively

Strain DNA content relative to haploid strain S288c

FY1679 2.00a ± 0.00

HA 1841 3.01b ± 0.08

HA 1842 3.07b ± 0.07

VIN7 3.04b ± 0.08

SOY3 2.89b ± 0.09

CECT 1388 3.25b ± 0.09

CECT 1990 2.86b ± 0.07

CECT 11002 3.02b ± 0.14

CECT 11003 3.21b ± 0.09

CECT 11004 3.13b ± 0.07

CECT 11011 2.99b ± 0.05

W27 3.18b ± 0.08

W46 3.20b ± 0.07

441 3.10b ± 0.09

SPG16-91 3.14b ± 0.08

PB7 3.96c ± 0.08

AMH 3.85c ± 0.18

Results are the mean value of three replicates. Means with the same letters donot differ significantly by one way ANOVA and Tukey’s HSD tests (p < 0.05).

ResultsHybrid genome structuresCaryoscopes, representing log2 hybridization ratios foreach gene mapped onto its corresponding chromosomeposition, of six hybrid strains from wine and 6 hybridsfrom brewing were obtained by array comparative gen-omic hybridization (aCGH) (Additional file 3: Figure S2).Due to the normalization procedure, the hybridizationratios derived from aCGH analysis show the relative pro-portions of each gene with respect to the average num-ber of copies in the hybrid, allowing the identification ofover- and underrepresented regions in the hybrid gen-ome. However, aCGH analysis in combination withploidy estimates and with information on the presence/absence of the chromosome regions coming from eachparental species, obtained in a previous PCR-RFLP ana-lysis of these hybrids [2,19], allowed us to decipher thegenome composition of hybrids.This way, ploidy estimates for these hybrids were

obtained by flow cytometry. The initial estimates with thepropidium iodide method suggested that most hybrids werediploids or close to diploidy (relative C-values of 2.0 to 2.6).However, these ploidy values were not congruent with thecaryoscope and PCR-RFLP data. The ratio-basednormalization of hybridization signals adjusts the averagesignal ratios (problem strain/reference strain) to 1, andhence the log2 values to 0. In the analysis of hybrids, ploidyestimates were 2n-2.6n, corresponding on average to a sub-genome coming from each parental species, i.e. for eachgene there are on average a copy coming from S. cerevisiaeand another from S. kudriavzevii. Due to the high astrin-gent hybridization conditions used in the aCGH analysis ofhybrids, only the S. cerevisiae subgenome is hybridizing, asconfirmed by the negative control performed with S.kudriavzevii DNA. Therefore, in the normalization ofhybridization signals, these ratios correspond to the adjust-ment of average signals coming from 1 S. cerevisiae genecopy from the hybrid to 1 gene copy form the referencehaploid S. cerevisiae strain. In the case of an increase ofcopy numbers in specific genes or chromosomal regions,log2 values should be higher than 0 (1, 2, etc. depending onthe number of copies), but in the case of loss of S. cerevisiaegene copies in the hybrid, a ratio of 0 (log2 of – ∞) shouldbe observed. However, 3–4 levels of log2 values, includingnegative but not infinite, are observed for some hybrids(Additional file 3: Figure S2), which made difficult the

interpretation of the aCGH results and suggested thatploidy estimates with propidium iodide were wrong.Therefore, new ploidy estimates of hybrids were

obtained by using SYTOX Green as the DNA-bindingdye, because Haase and Reed [14] demonstrated thatimproves linearity between DNA content and fluores-cence, and decreases peak drift associated with changesin dye concentration, growth conditions or cell size. Inthis new ploidy analysis, Swiss wine hybrids analyzed inour previous study [13] were also included.The statistical analysis of the new estimates showed

two significantly different groups of hybrids according toploidy levels: most hybrids, including the Swiss winestrains, appear as allotriploids and hybrids AMH andPB7 as allotetraploid yeasts (Table 2). The new ploidyestimates are in agreement with the different levels ofhybridization observed in the aCGH analyses and alsowith the previous PCR-RFLP analysis of hybrids [19].Final genome compositions were inferred for all

hybrids as depicted in Figure 1, taking into considerationflow cytometry data, aCGH and PCR-RFLP data. For ex-ample, in the case of the partial allotetraploid hybridAMH, its caryoscope shows four different hybridizationlevels, which correspond to 2 copies of S. cerevisiaegenes located in chromosomes (chr.) I and VI; 3 copiesof S. cerevisiae genes located in chr. VIII, IX and XIII,short chr. IV left region, chr. VII left arm, and chr. XV

I

II

100%

100%

Vin7AMH HA 1841 HA 1842 PB7 SOY3

I

III I

II100% 100%

100%100%

II

III

IV100%

III

IV

II

III

IV

II

III

IV

II

III

II

III

9%

100%

100%

100%

100%

100%

100%

100%

100%

100%

100%

*

V

VI 100%

100%V

VI

V

VI

V

VI

IV

V

VI

IV

V

VI100%100% 100%

100%

100%

100%

100%

100%

100%

100%

VII

VIII

67%

100%

VII

VIII

VII

VIII

VII

VIII

VII

VIII

VII

VIII

58% 99%

100%

99%

100%

99%

100%

99%

100%

*

IX

X

100%

100%

IX

X

IX

X

XI

IX

X

IX

X

IX

X

XI

100% 100%

100%

100%

100%

100%

100%

100%

100%

100%

100%

XI

XII

100%

100%

XI

XII

XII

XIII

XI

XII

XIII

XI

XII

XI

XII

XIII

100%

100%

100%

100%

100%

94%

100%

100%

100%

100%

XIII

XIV

100%

100%

XIII

XIV

XIV

XV

XIII

XIV

XV

XIII

XIV

XIV

XV

98% 100%

100%

100%

100%

100%

100%

100%

100%

100%

XV

XVI

100%

100%

XV

XVI

XVIXVI XV

XVI

XVI79%

100%98% 100%

98%

100%

I

CECT 1388

I

CECT 1990

I

CECT 11002

I

CECT 11003CECT 11004 CECT 11011

100%100%

100%I

II

III

I

II

III

I

II

III

II

III

100%

100%

100%

100%

100%

23%

100%

100%

100%

I

II

III

100%

100%

IV

V

IV

V

IV

V

IV

V

100%

100%

100%

92%

100% 79%

19%

100%

IV

V 100%

100%

VI

VII

VIII

VI

VII

VI

VII

VI

VII

95%

100%

100%

100%

100%

94%

100%

92%

VI

VII

100%

92%

100%

IX

X

VIII

IX

X

VIII

IX

X

VIII

IX

100%

100%

100%

100%

100%

98%

51%

100%

46%

VIII

IX

X

100%

100%

XI

XII

XI

XII

XI

XII

X

XI

XII

100%

100%

100%

100%

100%

76%

100%

100%

100%

XI

XII

100%

100%

XIII

XIV

XV

XIII

XIV

XIII

XIV

XIII

XIV

87%

100%

100% 26%

87%

%

100%

97%

XIII

XIV

XV

100%

100%

XV

XVI

XV

XVI

XV

XVI

XV

XVI

%

98%

100%

98%

100%

98%

79%

100%

XVI

100%

100%

Figure 1 (See legend on next page.)

Peris et al. BMC Genomics 2012, 13:407 Page 5 of 14http://www.biomedcentral.com/1471-2164/13/407

(See figure on previous page.)Figure 1 Genome composition of hybrids deduced from aCGH analysis, ploidy estimates and a previous analysis of absence/presenceof parental genes by RFLP analysis [2,19]. White and black bars are used to represent the S. cerevisiae and S. kudriavzevii genome fractions,respectively. Chromosomes showing black and white regions correspond to chimerical chromosomes. The percentages of S. kudriavzevii genesmaintained in each chromosome are shown for each chormosome. Strains names are depicted on a black or a gray background correspondingto wine or brewing strains, respectively. Asterisks in AMH Chr. III and VII indicate regions where non-reciprocal translocations or segmentalduplications can be present.

Peris et al. BMC Genomics 2012, 13:407 Page 6 of 14http://www.biomedcentral.com/1471-2164/13/407

right arm; 4 copies of S. cerevisiae genes located in chr.II, III left, IV right, V, VII right, X, XI, XII, XIV, XV left,and XVI; and 5 copies of S. cerevisiae genes located inchr. III left region and in a segment of chr. VII (Figure 1;Additional file 3: Figure S2).According to this combined analysis, 11 different pat-

terns were differentiated in the 12 hybrids under analysis.As a general rule, different degrees of loss of S. kudriavze-vii gene content in most hybrids were observed. Only theallotetraploid hybrid PB7 maintains a complete diploid setof chromosomes from each parental species, with the ex-ception of a small segment located in the left arm ofchromosome XI of the S. kudriavzevii subgenome. On thecontrary, the largest reduction of the S. kudriavzevii genecontent is observed in the partial allotetraploid hybridAMH, which lost 72% of the S. kudriavzevii genes. Therest of hybrids, all of them allotriploid, showed intermedi-ate situations derived from ancestors containing a diploidset of S. cerevisiae chromosomes and a haploid set of S.kudriavzevii chromosomes.These combined analyses also allowed us to detect dif-

ferent types of chromosome rearrangements present inhybrids: i) the complete loss of a S. kudriavzevii parentalchromosome compensated by an extra copy of theS. cerevisiae chromosome (chr. II,,III, V, X, XI, XII, XIVand XVI in AMH; chr. V in HA1841; chr. IV, IX and XIIin CECT 11002; chr. I in CECT 11011); ii) aneuploidies(chr. I, VI and VIII in AMH; chr. IX in CECT 1388; chr.XIV in CECT 1990; chr. IX in CECT 11002; chr. III andV in CECT 11003 and CECT 11004; chr. III in VIN7),and iii) the presence of chimerical chromosomes (chr.IV, VII and XV in AMH; chr. XI in PB7; chr. IV inSOY3; chr. VII in VIN7; chr. VII and XIV in CECT1388; chr. IV and XVI CECT 1990; chr. II, V, VII, X, XI,XIII and XIV in CECT 11002; chr. IV, V, VII, IX, XIVand XV in CECT 11003 and CECT 11004; and chr. VIIin CECT 11011); (see Figure 1).These chimerical chromosomes are characterized by

over- and underrepresented regions evidenced as up anddown jumps in the log2 ratio in the caryoscopes, whichare indicative of probable non-reciprocal recombinationevents between homeologous chromosomes (homologousfrom different species) (Table 3). The recombination sitesin the chimerical chromosomes were mapped accordingto the genome browser from Saccharomyces genomedatabase (SGD). Using a windows size of 15–20 Kb (four

genes in the left and right of the most plausible recombin-ation point) we found Ty elements, ARS sequences, clus-ters of homologous regions (CHRs) and tRNA elementsthat may have facilitated the recombination of the twohomologous parental chromosomes (Table 3). In severalcases, a common recombination site was observed inchromosomes belonging to two or more hybrids, indica-tive of common ancestry. This is the case of chromosomesIV, V, IX, XIV and XV in brewing hybrids CECT 11003and 11004; chromosome XIV in CECT 1388 and 11002;chromosome XV in CECT 11003, 11004 and AMH andchromosome VII in hybrids CECT 11003, CECT 11004,CECT 11002, CECT 11011 and CECT 1388 (Table 3 andAdditional file 3: Figure S2).

S. cerevisiae gene depletions in hybridsAlthough hybrids maintain in their genomes at least acomplete set of S. cerevisiae chromosomes, aCGH datafrom all hybrids analyzed in this work, as well as fromthose previously analyzed [13], can be used to determinethe common fraction of S. cerevisiae genes showing genecopy variations in hybrids compared to the referencestrain S288c. A common set of genes showing the samecopy number variations in hybrids may be indicative ofcommon origins.The analysis of the S. cerevisiae gene content from all

hybrids revealed the presence of less copies of a com-mon set of genes. Among them, the most interestingwere CUP1, ASP3, and ENA gene families, as well as Tyelements and 13 ORFs of unknown function (Additionalfile 4: Table S2). In general, copy variations in the S. cer-evisiae genome fraction of the hybrids were found ingenes located in subtelomeric regions (Additional file 3:Figure S2), although in some cases involve genes locatedin intrachromosomal regions, such as CUP1.Short segment amplifications were also detected in the

aCGH analysis. This was the case of hybrid AMH thatshowed three short region amplifications in chr. III, VIIand XIII. The higher hybridization signals of geneslocated in the two first regions could be postulated as in-dicative of the presence of chimerical chromosomes,however according to the previous PCR-RFLP analysisS. kudriavzevii genes were absent. Other amplificationsof S. cerevisiae segments located in chromosome XVIare observed in hybrids CECT 1388 (between genesYPL159C and YPL126W) and CECT 11002 (between

Peris et al. BMC Genomics 2012, 13:407 Page 7 of 14http://www.biomedcentral.com/1471-2164/13/407

YPL141C and YPL126W). Finally, a deleted region wasfound in one of the two copies of S. cerevisiae chromo-some XIV from strain CECT 1990 (between lociYNR013C and YNR031C) (Additional file 3: Figure S2).

S. kudriavzevii gene content and Gene Ontology (GO)analysesData obtained from all hybrids analyzed in this work aswell as from those previously analyzed [13] were alsoused to evaluate the presence of common S. kudriavzeviigenes (Additional file 5: Table S3). These common set ofgenes could be interesting to unveil potentially genes ofadaptive value in hybrids.As a general rule, most hybrids maintained around 90%

of the S. kudriavzevii genome, with the exception of thebrewing strain CECT 11002 and the wine strain AMHwhich only maintain 56.9% and 30.5% respectively.To determine if a group of S. kudriavzevii genes asso-

ciated with particular cellular components, molecularfunctions or biological processes may have been main-tained in all hybrids due to potential adaptive value, fourdifferent gene ontology (GO) term enrichment analyseswere performed (Additional file 6: Table S4). The firstanalysis included all wine and brewing hybrids. Due tothe low representation of the S. kudriavzevii genomefraction in AMH, this strain was removed from this firstanalysis. Gene ontology analysis was also separately per-formed according to the source of isolation of hybrids,wine and brewing fermentations. GO terms showing sig-nificant values were sorted according to their corre-sponding GO categories (Additional file 6: Table S4).Table 4 shows only those significantly represented GOterms of putative importance for wine or brewingfermentations.Significantly represented GO terms common to both

wine and brewing hybrids mainly corresponded to genesrelated to fatty acid metabolism (particularly transport),sulfur metabolism and the NAD+ salvage pathway.Genes associated with amino acid metabolism (N-linkedglycosylation and glutamate metabolism) were alsorepresented (Table 4).GO terms related to amino acid N-linked glycosilation

were also significantly present in hybrids from wine andbrewing analyzed independently. Moreover, GO termsassociated with ergosterol biosynthesis and mitochon-drial transport were also significantly detected in winehybrids; while those related to metabolism of amino acidssuch as glycine, threonine, arginine and proline, sulfurmetabolism, as well as fatty acid elongation were signifi-cant present in brewing strains (Table 4). Finally, an inde-pendent analysis of significant GO terms for AMHhybrid revealed the presence of genes involved in hyper-osmotic response, glycerol-3-phosphate dehydrogenase

complex, histidine biosynthesis and fatty acid metabolism(Table 4).

Phylogenetic relationships among hybridsA maximum parsimony tree was constructed based inpresence/absence of chromosomes and chromosomeregions data obtained for each particular genetic eventin all analyzed hybrids. The tree topology revealed thepresence of two main groups containing most allotri-ploid hybrids, particularly those from wine (Figure 2).Group I was constituted by Swiss wine strains W46,

441, W27 and SPG 16-91 as well as the brewing strainsCECT 11003 and CECT 11004. This group is supportedby the presence of five shared chimerical chromosomesas well as the CYC3 K2 allele [2].Group II includes the remaining allotriploid wine

hybrids HA1841, HA 1842, VIN7 and SOY3. This groupis only supported by the common presence of S. kudriav-zevii K2 alleles for genes EUG1 and APM3 [19], and thepossession of a higher fraction of S. kudriavzevii genome.The rest of the allotriploid hybrids, isolated from

brewing, and the wine allotetraploid PB7 and AMHstrains, appeared in separated branches with strain-specific chromosomal rearrangements. The only excep-tion is the shared loss of S. kudriavzevii chr. XII betweenthe partial allotetraploid AMH and the allotriploidCECT 11002, which can be considered a convergentevent. PB7 also shared similar restriction alleles withGroup II but this strain is also allotetraploid (Table 2).This most parsimonious tree shows several convergent

events, such as chromosomal losses, chromosomal rear-rangements and restriction site changes (evidencing dif-ferent allelic variants). S. kudriavzevii chr. I seems tohave been lost independently in hybrids SPG 441, CECT11011, and AMH. In a similar way, the lack of chr. V inhybrids HA 1841 and AMH, and chromosome XIV inCECT 1990 and AMH seem to be independent eventsaccording to this parsimony analysis.Convergent events involving recombinant chromo-

somes were also found. This is the case of the type 2 re-combination in chr. IV (shared by AMH and SOY3),type 1 recombination in chr. VII (shared by CECT11002, CECT 11011, 1388, W46, CECT 11003 andCECT 11004), type 2 recombination in chr. XIV (CECT1388 and CECT 11002) and the recombinant chr. XV(AMH and Group I hybrids). This could be indicative ofthe presence of recombination hotspots in the Saccharo-myces genomes.

DiscussionThe genome diversity in S. cerevisiae × S. kudriavzeviihybridsThe genome composition of 11 new wine and brewingS. cerevisiae x S. kudriavzevii hybrid strains was

Table 3 List of chimerical chromosome (CC) types found in the different S. cerevisiae × S. kudriavzevii hybrids

Chr. CC type Strains Breakpoint mappinginterval

Putative recombiningsequences

II type 1 CECT 11002 YBL018C-YBL011W Ty1 LTR, Ty3 LTR, tRNA-Ile,tRNA-Gly, ARS

IV type 1 W27, W46, 441, SPG16-91,CECT 11003, CECT 11004

YDL095W PMT1 (ref. [13])

type 2 AMH, SOY3 YDL185W-YDL179W CHR 12

type 3 CECT 1990 YDL185W-YDL179W CHR 12

V type 1 W27, W46, 441, SPG16-91,CECT 11003, CECT 11004

YER006W NUG1 (ref. [13])

type 2 CECT 11002 YEL018C-YEL011W Ty1 LTR, Ty4 LTR, tRNA-Gln

VII type 1 W46, CECT 11003, CECT 11004,CECT 11002, CECT 11011, CECT 1388

YGR249W-YGR244C ARS, CHR 29

type 2 AMH YGR062C-YGR058W CHR 30

type 3 VIN7 YGR106C-YGR112C tRNA-Leu, tRNA-Lys, Ty1 LTR,tRNA-Cys, Ty3 LTR, ARS

IX type 1 W27, W46, 441, SPG16-91,CECT 11003, CECT 11004

YIL053W RHR2-RPL34B (ref. [13])

X type 1 CECT 11002 YJL039C-YJL036C tRNA-Asp, tRNA-Arg, Ty1 LTR,ARS, tRNA-Val

XI type 1 CECT 11002 YKR025C-YKR028W Ty1 LTR

type 2 PB7 YKL203C-YKL204W ARS

XIII type 1 CECT 11002 YML012C-YML009W-B CEN13, ARS

XIV type 1 W27, W46, 441, SPG16-91,CECT 11003, CECT 11004

YNR001C CEN14 (ref. [13])

type 2 CECT 1388, CECT 11002 YNR029C-YNR032W ARS

XV type 1 W27, W46, 441, SPG16-91,CECT 11003, CECT 11004, AMH

YOL053W THI20-PSH1 (ref. [13])

XVI type 1 CECT 1990 YPR007C-YPR011C Ty1LTR, tRNA-Gly, tRNA-Lys,

Chr., chromosome number; CHR, cluster of homology region. Strain names in italics correspond to wine hybrids and in bold to brewing hybrids. Somerecombination sites were described elsewhere [13], as indicated.

Peris et al. BMC Genomics 2012, 13:407 Page 8 of 14http://www.biomedcentral.com/1471-2164/13/407

described in this work by means of aCGH analysis. Add-itionally, a comparison between them and other fourwine hybrids already described by [13] was also per-formed. Individual and differential chromosomal com-position patterns were found for each particular strain,except for brewing strains CECT 11003 and CECT11004 which appear closely related to the previouslydescribed Swiss wine hybrids [13]. The close relation-ships between wine hybrid strains from Switzerland andthe brewing strains CECT 11003 and 11004 was alreadyobserved in a previous study based on PCR-RFLP ana-lysis of hybrids as well as in the phylogenetic recon-struction based on COX2 sequences [19]. In that work,a recombination in chromosome XV was proposed asthe unique difference between strains 11003 and 11004;however, aCGH analysis carried out in this studydemonstrated that this recombination is present in bothstrains (Figure 2). These Swiss wine hybrids were previ-ously described as diploids [13] on the basis of ploidyestimations with propidium iodide. However, in the

reanalysis of ploidy with SYTOX Green, they alsoresulted to be allotriploids as CECT 11003 and CECT11004.Flow cytometry results with SYTOX Green were in ac-

cordance with genome structure deduced from aCGHanalysis carried out in this work and with the presence/ab-sence of parental genes deduced from a previous PCR-RFLP analysis of hybrids [19]. Most S. cerevisiae × S.kudriavzevii hybrid strains were allotriploids, with the ex-ception of AMH and PB7 which were allotetraploids.Some aneuploidies were also found in several hybrids.Aneuploidies seem to be common in Saccharomyceshybrids since this phenomenon have also been observedin S. cerevisiae × S. bayanus hybrids [22,23]. The role ofaneuploidies in the hybrid genomes is not clear, but theirpresence in S. cerevisiae affected both the transcriptomeand proteome, generating significant phenotypic variationand bringing fitness gains under diverse conditions [24].Recently, the hybrid genome of VIN7, one the hybrids

analyzed in the present study, has completely been

Table 4 Summary of the most relevant metabolic pathways and biological processes obtained after Gene Ontologyanalysis using the S. kudriavzevii genes retained in each group of hybrids

Group of hybrids GO ID GO Term Npresent/Nmeasured % p-value

WINE 6487 Protein amino acid N-linked glycosylation 36/42 85.7 0.013

6839 Mitochondrial transport 10/10 100 0.033

Ergosterol Biosynthesis 17/19 89.5 0.049

BREWING 6487 Protein amino acid N-linked glycosylation 28/42 66.7 0.017

Fatty acid elongation saturated 4/4 100 0.039

Glycine serine and threonine metabolism 27/42 64.3 0.03

Arginine_and_proline_metabolism 16/23 69.6 0.049

Sulfur_Degradation 4/4 100 0.048

ALL 6487 Protein amino acid N-linked glycosylation 25/42 59.5 0.003

15908 Fatty acid transport 4/4 100 0.025

Glutamate metabolism 15/27 55.6 0.046

Sulfur metabolism 8/11 72.7 0.021

NAD salvage pathway 5/6 83.3 0.027

Sulfate assimilation pathway II 5/6 83.3 0.019

AMH 6972 Hyperosmotic response 5/7 71.4 0.036

9331 Glycerol 3 phosphate dehydrogenase complex 3/3 100 0.033

Histidine biosynthesis 5/7 71.4 0.039

Fatty acid metabolism 11/17 64.7 0.010

Due to the massive S. kudriavzevii gene losses in AMH, this strain was not included in any grouping, and hence, analyzed alone.

Peris et al. BMC Genomics 2012, 13:407 Page 9 of 14http://www.biomedcentral.com/1471-2164/13/407

sequenced [25], concluding that this strain is an almostperfect allotriploid hybrid that contains a heterozygousdiploid S. cerevisiae genome and a haploid S. kudriavze-vii genome. The genome constitution of VIN7 deducedfrom the sequencing analysis is basically similar to theone inferred by aCGH in the present study, but thereare some differences. The genome sequence analysisdetected a homeologous recombination generating a chi-merical chromosome VII, a genomic substitution of a re-gion of 15 kb, of S. kudriavzevii genomic DNA fromchromosome IV by the orthologous sequences fromS. cerevisiae and a genomic substitution of a 13 kb re-gion of S. cerevisiae genomic DNA from chromosomeIV by S. kudriavzevii sequences combined with homeo-logous recombination between the S. kudriavzevii andS. cerevisiae alleles. The first rearrangement involving achimerical chromosome VII was clearly detected in theaCGH analysis, but not the two genomic substitutions.Both genomic substitution involve short segmentalreplacements of a few genes (7 and 8), and the secondan almost reciprocal recombination between homeolo-gous chromosomes that cannot be observed by aCGHanalysis. However, the presence/absence analysis ofparental genes in hybrids [19] detected the loss ofS. kudriavzevii chromosome III in our VIN7. As an on-going project, our group is also sequencing the wholegenome of several S. cerevisae × S. kudriavzevii hybrids,including the commercial VIN7 yeast. We checked in

the preliminary sequencing of our VIN7 strain for thepresence of S. kudriavzevii chromosome III sequencesand the result was negative, confirming our aCGHresults and indicating that our VIN7 strain is different.These differences may be due to the fact that our VIN7strain was isolated from a commercial dry yeast sampleprovided by Anchor Yeast but Borneman et al. [25]sequenced the original mother culture of VIN7, as theymention in their acknowledgements. Therefore, the con-tinuous propagation of this yeast in molasses under aer-obic conditions to obtain commercial dry yeasts mayhave promoted a new chromosomal rearrangement, theloss of the S. kudriavzevii chromosome III.Taking into consideration the ploidy data as well the fact

that most hybrids possess either trisomic (2 S. cerevisiaechromosomes: 1 S. kudriavzevii chromosome) or tetraso-mic chromosomes (2 S. cerevisiae chromosomes: 2 S.kudriavzevii chromosomes), two scenarios on thehybridization process are plausible. In the case of allotri-ploid hybrids, the simplest explanations for their originsare hybridization events by rare-mating between a diploidcell of S. cerevisiae and a haploid cell or spore of S.kudriavzevii. This is also supported by the genome se-quencing of VIN7, one of the allotriploid strains, whichresulted to contain heterozygous diploid genome from S.cerevisiae and a haploid genome from S. kudriavzevii [25].On the other hand, diploid and diploid cell rare-

mating between S. cerevisiae and S. kudriavzevii should

1990

1388

rec1 VII

1100311004

PEX

2 C

2

rec1 XII

rec3 IV

EUG1 K2

APM3K2

11011re

c1 V

II

rec

Xku

d IX

loss

rec2

XIV

rec1

VII

kud

I lo

ss

rec3 VII

kud III loss

16-91

rec1

XI

APM

3 K

2

rec

II

rec2 XI

11002

rec2

V

rec

XII

I Group IEUG1 K2

APM3 K2VIN71842

PB7

kud V losskud

XII

loss

1841

Group IISOY3

AMH

441W46

W27

Figure 2 Maximum parsimony tree indicating the minimum number of chromosomal rearrangements and restriction site changes(presence/absence matrix is given in Additional file 2: Table S1) necessary to connect the different genotypes exhibited by the S.cerevisiae × S. kudriavzevii hybrids to a putative hybrid ancestor. This putative ancestor is not necessarily the same for all lineages, it justcorresponds to an ancestral state containing the complete S. cerevisiae and S. kudriavzevii genomes, but it could be generated several times fromdifferent parental strains, as discussed in the main text. Genotypes are represented by white and gray circles for wine and brewing hybrids,respectively. Rearrangements are indicated by arrows giving the direction of the irreversible change and were treated under the Camin-Sokalcriterion. Rearrangements were assumed to be caused by nonreciprocal recombination (rec) among homoeologous chromosomes (romannumbers) and whole chromosome losses (loss) of one of the parental chromosomes (kud, S. kudriavzevii). Restriction site changes can bereversible (gains/losses represented by diamonds) and were treated under the Wagner criterion. The gene region and the restriction patternsinvolved are also indicated (for a description see references [2] and [19]).

Peris et al. BMC Genomics 2012, 13:407 Page 10 of 14http://www.biomedcentral.com/1471-2164/13/407

be invoked to explain the origin of allotetraploid hybrids.In the case of PB7 it was observed high spore viability(95%) due to the presence of the two chromosomes cop-ies of each parental strain.Rare-mating between diploid cells was already pro-

posed as a probable mechanism for hybrids generation[13,26]. However, haploid cell or spore mating betweenS. cerevisiae and S. kudriavzevii, followed by whole gen-ome duplications due to endoreplication or chromosomeduplications due to non-disjunction, and subsequentchromosomal rearrangements, although less plausible,cannot totally be discarded.

Characterization of the S. kudriavzevii subgenome fromhybridsAccording to Sipiczki [26], genomes from each parentalspecies interact in the new hybrid genome. This inter-action can be observed in the loss of large parts of oneor both genomes as well as in the presence of chimericalchromosomes that make the hybrid genome as stable aspossible to future genetic modifications. Additionally,adaptive evolution of these hybrid genomes under fer-mentative environmental conditions could make hybrid

genome to conserve the chromosomes, or part of them,which grant a selective advantage [27]. According to theresults obtained in this work as well as in our previousstudies [2,13,19], S. cerevisiae × S. kudriavzevii hybridsseem to have the common trend to lose the S. kudriave-vii parental chromosomes maintaining the S. cerevisiaeones. The reduction of the non-S. cerevisiae genomeobserved in both wine and brewing S. cerevisiae × S.kudriavzevii hybrids was already reported for artificial S.cerevisiae × S. uvarum hybrids genetically stabilized bysuccessive sporulation steps [28]. In contrast, S. pastor-ianus (S. cerevisiae × S. eubayanus hybrids) Group 1strains obtained from different brewing processes andstudied by aCGH analysis, showed a trend to lose the S.cerevisiae genome fraction [22]. The cause of the pre-dominance of one or the other parental genome in thehybrids remains unclear yet. However, selective pres-sures acting under harsh environmental conditions andcytonuclear interactions have been suggested as themain factors affecting the genome conformation ofhybrids. In S. cerevisiae × S. eubayanus lager strains,supposed to be naturally selected after years of use inbrewing, the predominance of a S. eubayanus-like

Peris et al. BMC Genomics 2012, 13:407 Page 11 of 14http://www.biomedcentral.com/1471-2164/13/407

genome has been related to the maintenance of the S.eubayanus mitochondria [22,29]. However, artificialhybrids constructed from the same two parental species,but without selective pressures, inherited their mito-chondrial genome from either one or the other parentalspecies randomly [29,30]. The conservation of the mito-chondrial genome from the parental species most repre-sented in the nuclear genome was also observed in thestable artificial S. cerevisiae × S. uvarum hybrids, whichmaintained the mitochondrial genome of the S. cerevi-siae parental strain [28]. All S. cerevisiae × S. kudriavze-vii natural hybrids analyzed in this work, except forAMH, maintained a S. kudriavzevii mitochondrial gen-ome [2,19]. However, S. cerevisiae × S. kudriavzevii arti-ficial hybrids, randomly inherited the S. cerevisiae or theS. kudriavzevii mitochondrial DNA (Pérez-Través et al.personal communication). This discrepancy between themtDNA inheritance in artificial vs. natural hybrids hasbeen associated with the result of an unwitting human-driven selection of naturally generated hybrid strains forfermentations at low temperature [29]. A common ori-gin for all hybrids could be another possible explanation,but the present analysis of the genome constitutions inhybrids suggests diverse origins.Interestingly, the hybrid AMH, which maintained the

S. cerevisiae mitochondria, has lost a 69% of the nucleargenes of S. kudriavzevii coding for proteins with func-tions associated to the mitochondria; while the rest ofthe analyzed hybrids with S. kudriavzevii mitochondriahave lost only 0.67%–42.48% of the S. kudriavzevii genesrelated to mitochondrial functions. Due to the fact that anumber of mitochondrial proteins encoded in the nu-clear genome play an important role in the mtDNA repli-cation and transmission, both the type of mitochondrialDNAs and the functions of the mitochondria in a hybridstrain are clearly under the control of the nuclear genome[31]. One of the most interesting evidence about nuclear-mitochondrial genome interactions were described by Leeet al. [32], who demonstrated that the presence of the S.bayanus nuclear gene AEP2 together with the S. cerevisiaemitochondrial gene OLI1 cause a cytonuclear incompati-bility. More recently, Chou et al. [33] identified other twogenes, MRS1 and AIM22, associated with cytonuclear in-compatibility among S. cerevisiae, S. paradoxus and S.bayanus. A similar behavior involving the same or otherdifferent genes in S. cerevisiae × S. kudriavzevii hybridswas not yet demonstrated.aCGH and GO analysis carried out with those S.

kudriavzevii genes conserved in all S. cerevisiae × S.kudriavzevii hybrids with S. kudriavzevii mitochondria(excluding AMH) evidenced a significant enrichment innuclear genes related to mitochondrial function (a totalof 328 genes) supporting the hypothesis of a necessaryinteraction between the S. kudriavzevii nuclear-encoded

proteins and the mitochondrial genomes or their pro-ducts. Taking into consideration that a total of 751 pro-teins encoded by the nuclear genome are associated withthe mitochondrial function in S. cerevisiae [34], and con-sidering a similar number in S. kudriavzevii, we can as-sume that the remaining genes up to 751 might be non-essential for the maintenance of the S. kudriavzeviimitochondria in hybrids. In particular the S. kudriavzeviigene AEP2 reported by Lee et al. [32] was not commonto all analyzed hybrids, indicating that different incom-patible nuclear-mitochondrial pair of genes could beassociated with each particular pair of Saccharomycesparental species involved in hybrid generation.GO analysis was also very informative with regards to

the conservation in hybrids of particular groups of genes,inherited from each parental species, that may be poten-tially related to adaptive advantage for fermentation at lowtemperatures. A significant overrepresentation of S.kudriavzevii genes associated with the physiological adap-tation of yeasts to grow at low temperatures, such as fattyacid transport and N-glycosilation of proteins in allhybrids, and ergosterol biosynthesis in the case of winehybrids [35-37] was observed (Table 4). Changes in mem-brane fluidity are the primary signal triggering the coldshock response [35]. This response involves certain groupsof genes: members of the DAN/TIR family of cell-wallmannoproteins, genes coding for temperature inducibleprotein (TIP1) and seripauperins (PAU), genes related toergosterol and phospholipid synthesis (ERG, INO1 andOPI3) and the gene coding for the only known desaturasein S. cerevisiae (OLE1), among others [35]. These sets ofgenes are present in the S. kudriavzevii subgenome of allhybrids analyzed in this work, with some exceptionsmainly involving AMH (Table 4 and Additional file 6:Table S4).Our results are in agreement with results about stress

olerance, including adaptation to low temperatures, previ-ously obtained in our laboratory using some of the S. cere-visiae × S. kudriavzevii hybrids analyzed in this work[11,38]. Physiological implications of possessing S.kudriavzevii genes in those particular functions or meta-bolic pathways must be elucidated in future studies involv-ing both transcriptomic and metabolomic analyses.

Wine yeast signatures in the S. cerevisiae subgenomefrom hybridsAn interesting result obtained from aCGH analysis wasthe detection of a common set of S. cerevisiae genes thatare in lower copies in the genome of all S. cerevisiae × S.kudriavzevii hybrids (Additional file 4: Table S2). Thisfinding might indicate that the S. cerevisiae parentalstrains involved in the different hybridization eventsshared a similar genetic background and were closelyrelated yeasts.

Peris et al. BMC Genomics 2012, 13:407 Page 12 of 14http://www.biomedcentral.com/1471-2164/13/407

Using a similar methodology, a trend to loss some par-ticular set of genes in S. cerevisiae wine strains, withregards to strains belonging to the same species but iso-lated from different sources, was previously demon-strated [39,40]. Dunn et al. [39] proposed the term“commercial wine yeast signature” to refer to this set ofgenes. Most of these genes that are frequently depletedin wine strains are also depleted in the S. cerevisiae frac-tion of the hybrid genomes of all hybrids. This findingsupports the hypothesis that these hybrids have likelybeen generated from wine S. cerevisiae parental strains.

On the origin of hybridsThe maximum parsimony analysis of the relationshipsbetween the wine and beer hybrids are congruent withdiverse origins for the strains according to chromosomalrearrangement differences, mainly due to the presence ofchimerical chromosomes, and S. kudriavzevii chromo-some losses, in some cases compensated by the presenceof an extra copy of the homeologous S. cerevisiaechromosome (Figure 2).While the brewing strains seem to represent different

and divergent lines (except strains CECT11003 and11004), most wine hybrids clustered in two main groupsof strains sharing common events, with the exception ofAMH and PB7 that were independently originated.Brewing strains CECT 11003 and 11004 shared the samegenome than wine hybrid W46 probably evidencing thateither an original strain with this common genomestructure was introduced in both fermentative processes,or colonize one fermentative process from the other.The parsimony tree obtained in this study is congruentwith previous phylogenetic reconstructions of hybridsbased on COX2 sequences [19].The occurrence of several chimerical chromosomes

sharing similar—if not the same—recombination points,common to some S. cerevisiae × S. kudriavzevii hybridslocated in different branches of the parsimony tree, indi-cates the presence of recombination hot spots. Recombin-ation between homeologous chromosomes are probablymediated by highly recombining regions located in the re-combination sites, such as ARS sequences [41], Ty ele-ments [42], Y’ elements, rRNA regions and conservedcoding genes [13,43]. When recombination is initiated ina region with high homology, the mismatch repair system(MMR) stimulates the loss of one partner of the recom-bination event in the hybrids and the fixation of the other,thus generating a chimerical recombinant chromosome.With the exception of the almost perfect allotetraploidPB7, hybrids have low spore viability (<1%) indicating thatthey are maintained by mitotic budding. Therefore, mi-totic homeologous recombination, although much less fre-quent than meiotic, may also explain the generation ofchimerical chromosomes.

The genome composition of hybrids reveals that the an-cestral hybrid strains were allotriploid or allotetraploid,resulting from rare mating between diploid S. cerevisiaeand haploid or diploid S. kudriavzevii [4,25]. The presenceof triple hybrids also supports this hypothesis [1,19]. Fi-nally, the presence of S. kudriavzevii alleles shared be-tween most hybrids and the European S. kudriavzeviipopulation [10], as well as the presence of the gene GAL4from S. kudriavzevii [2,19], which is a functional gene inthe European populations of S. kudriavzevii but a pseudo-gene in the Japanese strains [44], indicate that thesehybrids were originated from a European S. kudriavzeviiparental strain.

ConclusionsHybridization between S. cerevisiae and S. kudriavzeviihave occurred several times by rare-mating between dif-ferent wine S. cerevisiae diploid and European S. kudriav-zevii haploid or diploid progenitors. After hybridization,the hybrid genome suffered random genomic rearrange-ments mediated by crossing-over between homologouschromosomes and non-disjunction, promoting the lossof variable fractions of the parental subgenomes. Boththe restrictions imposed byinteractions between bothparental genomes as well as between nuclear and mito-chondrial genomes, together with the selective environ-mental conditions prevailing during fermentationmodulated the final composition of the hybrid genomes,characterized by the maintaining of the S. cerevisiae gen-ome and the progressive reduction of the S. kudriavzeviicontribution.

Additional files

Additional file 1: Figure S1. Caryoscope representation of microarraydata of S.kudriavzevii IFO1802. Array CGH data are shown in numericalorder with chromosome I at the top and chromosome XVI at thebottom. Red signal indicates hybridization signal but it’s important tonote the high normalization factor applied to the red signal (2) and thecorrection applied to the green one (0.49). This figure indicate that nocross hybridization has occurred between S. kudriavzevii genes andS. cerevisiae genes.

Additional file 2: Table S1. Binary table showing the presence/absenceof a particular event for each hybrid strain (coded as 1/0 respectively).kud, S. kudriavzevii; loss, chromosome loss; rec, recombination generatinga chimerical chromosome. Roman numerals indicate chromosomenumbers and Arabic numerals the types of chimerical chromosomes orthe restriction sites in the analysis of RFLP patterns in 34 genes.Parsimony tree was computed using a mixture model in whichchimerical chromosomes, S. kudriavzevii chromosome losses and doubleS. cerevisiae chromosome were considered under the Camin-Sokalcriterion and restriction site gains/losses under the Wagner criterion.

Additional file 3: Figure S2. Caryoscope representation of microarraydata of 11 S. cerevisiae × S.kudriavzevii hybrids. Array CGH data are shown innumerical order with chromosome I at the top and chromosome XVI at thebottom for each strain. Regions with higher red signal correspond to S.cerevisiae genome regions that are overrepresented in the hybrid genomeand those with higher green signal to those regions that areunderrepresented. aCGH of wine and brewery hybrids are depicted on black

Peris et al. BMC Genomics 2012, 13:407 Page 13 of 14http://www.biomedcentral.com/1471-2164/13/407

and gray backgrounds, respectively. Arrows indicate potential non-reciprocalrecombination events between homeologous chromosomes involved inthe generation of chimerical chromosomes.

Additional file 4: Table S2. List of S. cerevisiae genes depleted in all S.cerevisiae × S. kudriavzevii hybrids under analysis.

Additional file 5: Table S3. S. kudriavzevii gene composition for eachhybrid. A cross indicates the putative presence of that gene byconsidering colinearity between S. kudriavzevii and S. cerevisiae genomes.R indicates interspecies recombinant genes. Wine and brewing hybridsare indicated in black and gray lettering, respectively.

Additional file 6: Table S4. Metabolic pathways and biologicalprocesses obtained from a Gene Ontology analysis using theS. kudriavzevii genes retained in each hybrid grouping.

Competing interestsThe author(s) declare that they have no competing interests.

Authors’ contributionsThis study is the result of the collaboration between AQ and EB laboratories.CB, AQ and EB conceived and supervised this study. DP, CB, AQ and EBdesigned the experiments. DP and CL performed the experimental work anddata analyses. DP and CB wrote the first version of the manuscript. CL, AQand EB participated in the final manuscript revision. All authors read andapproved the final manuscript.

AcknowledgementsThis work was supported by Spanish Governments projects AGL2009-12673-CO2-01 and AGL2009-12673-CO2-02 to AQ and EB respectively, and by grantPROMETEO/2009/019 from Generalitat Valenciana to AQ, EB and CB. DP andCL acknowledges to the Spanish Government for a FPI fellowship and apostdoctoral contract, respectively. We also thank Ksenija Lopandić, JoséManuel Álvarez-Pérez, Sandi Orlić, Lallemand Bio and Anchor Wine Yeasts forproviding yeast strains.

Author details1‘Cavanilles’ Institute of Biodiversity and Evolutionary Biology, University ofValencia, Parc Científic, P.O. Box 22085, E-46071, Valencia, Spain. 2Departmentof Biotechnology, Institute of Agrochemistry and Food Technology (CSIC),Valencia, Spain. 3Yeast Biodiversity & Biotechnology Group. InstitutoMultidisciplinario de Investigación y Desarrollo de la Patagonia Norte (IDEPA),CONICET, Universidad Nacional del Comahue, Neuquén, Argentina.

Received: 22 September 2011 Accepted: 4 April 2012Published: 20 August 2012

References1. González SS, Barrio E, Gafner J, Querol A: Natural hybrids from

Saccharomyces cerevisiae, Saccharomyces bayanus and Saccharomyceskudriavzevii in wine fermentations. FEMS Yeast Res 2006, 6:1221–1234.

2. González SS, Barrio E, Querol A: Molecular characterization of new naturalhybrids between Saccharomyces cerevisiae and Saccharomyceskudriavzevii from brewing. Appl Environ Microbiol 2008, 74:2314–2320.

3. Lopandić K, Gangl H, Wallner E, Tscheik G, Leitner G, Querol A, Borth N,Breitenbach M, Prillinger H, Tiefenbrunner W: Genetically different wineyeasts isolated from Austrian vine-growing regions influence winearoma differently and contain putative hybrids between Saccharomycescerevisiae and Saccharomyces kudriavzevii. FEMS Yeast Res 2007, 7:953–965.

4. de Barros Lopes M, Bellon JR, Shirley NJ, Ganter PF: Evidence for multipleinterspecific hybridization in Saccharomyces sensu stricto species. FEMSYeast Res 2002, 1:323–331.

5. Groth C, Hansen J, Piškur J: A natural chimeric yeast containing geneticmaterial from three species. Int J Syst Bacteriol 1999, 49:1933–1938.

6. Liti G, Peruffo A, James SA, Roberts IN, Louis EJ: Inferences of evolutionaryrelationships from a population survey of LTR-retrotransposons andtelomeric-associated sequences in the Saccharomyces sensu strictocomplex. Yeast 2005, 22:177–192.

7. Masneuf I, Hansen J, Groth C, Piškur J, Dubourdieu D: New hybridsbetween Saccharomyces sensu stricto yeast species found among wineand cider production strains. Appl Environ Microbiol 1998, 64:3887–3892.

8. Libkind D, Hittinger CT, Valério E, Gonçalves C, Dover J, Johnston M,Gonçalves P, Sampaio JP: Microbe domestication and the identification ofthe wild genetic stock of lager-brewing yeast. Proc Natl Acad Sci 2011,108:14539–14544.

9. Sampaio JP, Gonçalves P: Natural populations of Saccharomyceskudriavzevii in Portugal are associated with oak bark and sympatric withS. cerevisiae and S. paradoxus. Appl Environ Microbiol 2008, 74:2144–2152.

10. Lopes CA, Barrio E, Querol A: Natural hybrids of Saccharomyces cerevisiae ×Saccharomyces kudriavzevii share alleles with European wild populationsof S. kudriavzevii. FEMS Yeast Res 2010, 10:412–421.

11. Arroyo-López FN, Orlić S, Querol A, Barrio E: Effects of temperature, pHand sugar concentration on the growth parameters of Saccharomycescerevisiae, S. kudriavzevii and their interspecific hybrids. Int J FoodMicrobiol 2009, 131:120–127.

12. Salvadó Z, Arroyo-López FN, Barrio E, Querol A, Guillamón JM:Quantifying the individual effects of ethanol and temperature on thefitness advantage of Saccharomyces cerevisiae. Food Microbiol 2011,28:1155–1161.

13. Belloch C, Pérez-Torrado R, González SS, Pérez-Ortín JE, García-Martínez J,Querol A, Barrio E: Chimeric genomes of natural hybrids ofSaccharomyces cerevisiae and Saccharomyces kudriavzevii. Appl EnvironMicrobiol 2009, 75:2534–2544.

14. Haase SB, Reed SI: Improved flow cytometric analysis of the buddingyeast cell cycle. Cell Cycle 2002, 1:132–136.

15. Querol A, Barrio E, Ramón D: A comparative study of different methods ofyeast-strain characterization. Syst Appl Microbiol 1992, 15:439–446.

16. Brazma A: ArrayExpress, a public repository for microarray geneexpression data at the EBI. Nucleic Acids Res 2003, 31:68–71.

17. Myers CL, Dunham MJ, Kung SY, Troyanskaya OG: Accurate detection ofaneuploidies in array CGH and gene expression microarray data.Bioinformatics 2004, 20:3533–3543.

18. Cliften PF, Sudarsanam P, Desikan A, Fulton L, Fulton B, Majors J, WaterstonR, Cohen BA, Johnston M: Finding functional features in Saccharomycesgenomes by phylogenetic footprinting. Science 2003, 301:71–76.

19. Peris D, Belloch C, Lopandić K, Álvarez-Pérez JM, Querol A, Barrio E: Themolecular characterization of new types of S. cerevisiae × S. kudriavzeviihybrid yeasts unveils a high genetic diversity. Yeast 2012, 29:81–91.

20. Doniger SW, Salomonis N, Dahlquist KD, Vranizan K, Lawlor SC, Conklin BR:MAPPFinder: using Gene Ontology and GenMAPP to create a globalgene-expression profile from microarray data. Genome Biol 2003, 4:R7.

21. Felsenstein J: PHYLIP: Phylogeny Inference Package. v. 369th edition. Seattle:University of Washington; 2005.

22. Dunn B, Sherlock G: Reconstruction of the genome origins and evolutionof the hybrid lager yeast Saccharomyces pastorianus. Genome Res 2008,18:1610–1623.

23. Nakao Y, Kanamori T, Itoh T, Kodama Y, Rainieri S, Nakamura N, ShimonagaT, Hattori M, Ashikari T: Genome sequencing of the lager brewing yeast,an interspecies hybrid. DNA Res 2009, 16:115–129.

24. Pavelka N, Rancati G, Zhu J, Bradford WD, Saraf A, Florens L, Sanderson BW,Hattem GL, Li R: Aneuploidy confers quantitative proteome changes andphenotypic variation in budding yeast. Nature 2010, 468:321–325.

25. Borneman AR, Desany BA, Riches D, Affourtit JP, Forgan AH, Pretorius IS,Egholm M, Chambers PJ: The genome sequence of the wine yeast VIN7reveals an allotriploid hybrid genome with Saccharomyces cerevisiae andSaccharomyces kudriavzevii origins. FEMS Yeast Res 2012, 12:88–96.

26. Sipiczki M: Interspecies hybridisation and recombination inSaccharomyces wine yeasts. FEMS Yeast Res 2008, 8:996–1007.

27. Barrio E, González SS, Arias A, Belloch C, Querol A: Molecular mechanismsinvolved in the adaptive evolution of industrial yeasts. In Yeasts in Foodand Beverages. 1st edition. Edited by Querol A, Fleet GH. Berlin: Springer;2006:153–174.

28. Antunovics Z, Nguyen HV, Gaillardin C, Sipiczki M: Gradual genomestabilisation by progressive reduction of the Saccharomyces uvarumgenome in an interspecific hybrid with Saccharomyces cerevisiae. FEMSYeast Res 2005, 5:1141–1150.

29. Rainieri S, Kodama Y, Nakao Y, Pulvirenti A, Giudici P: The inheritance ofmitochondrial DNA in lager brewing strains. FEMS Yeast Res 2008, 8:586–596.

30. Solieri L, Antúnez O, Pérez-Ortín JE, Barrio E, Giudici P: Mitochondrialinheritance and fermentative: oxidative balance in hybrids betweenSaccharomyces cerevisiae and Saccharomyces uvarum. Yeast 2008,25:485–500.

Peris et al. BMC Genomics 2012, 13:407 Page 14 of 14http://www.biomedcentral.com/1471-2164/13/407

31. Vero LDE, Pulvirenti A, Gullo M, Bonatti P, Giudici P: Sorting ofmitochondrial DNA and proteins in the progeny of Saccharomycesinterspecific hybrids. Ann Microbiol 2003, 53:219–231.

32. Lee HY, Chou JY, Cheong L, Chang NH, Yang SY, Leu JY: Incompatibility ofnuclear and mitochondrial genomes causes hybrid sterility between twoyeast species. Cell 2008, 135:1065–1073.

33. Chou JY, Hung YS, Lin KH, Lee HY, Leu JY: Multiple molecular mechanismscause reproductive isolation between three yeast species. PLoS Biol 2010,8:e1000432.

34. Sickmann A, Reinders J, Wagner Y, Joppich C, Zahedi R, Meyer HE,Schönfisch B, Perschil I, Chacinska A, Guiard B, Rehling P, Pfanner N,Meisinger C: The proteome of Saccharomyces cerevisiae mitochondria.Proc Natl Acad Sci U S A 2003, 100:13207–13212.

35. Aguilera J, Rández-Gil F, Prieto JA: Cold response in Saccharomycescerevisiae: new functions for old mechanisms. FEMS Microbiol Rev 2007,31:327–341.

36. Beltrán G, Novo M, Leberre V, Sokol S, Labourdette D, Guillamón JM, Mas A,François J, Rozès N: Integration of transcriptomic and metabolic analysesfor understanding the global responses of low-temperature winemakingfermentations. FEMS Yeast Res 2006, 6:1167–1183.

37. Higgins VJ, Beckhouse AG, Oliver AD, Rogers PJ, Dawes IW: Yeast genome-wide expression analysis identifies a strong ergosterol and oxidativestress response during the initial stages of an industrial lagerfermentation. Appl Environ Microbiol 2003, 69:4777–4787.

38. Belloch C, Orlić S, Barrio E, Querol A: Fermentative stress adaptation ofhybrids within the Saccharomyces sensu stricto complex. Int J FoodMicrobiol 2008, 122:188–195.

39. Dunn B, Levine RP, Sherlock G: Microarray karyotyping of commercialwine yeast strains reveals shared, as well as unique, genomic signatures.BMC Genomics 2005, 6:53.

40. Carreto L, Eiriz M, Gomes A, Pereira P, Schuller D, Santos M: Comparativegenomics of wild type yeast strains unveils important genome diversity.BMC Genomics 2008, 9:524.

41. Di Rienzi SC, Collingwood D, Raghuraman MK, Brewer BJ: Fragile genomicsites are associated with origins of replication. Genome Biol Evol 2009,1:350–363.

42. Kim JM, Vanguri S, Boeke JD, Gabriel A, Voytas DF: Transposable elementsand genome organization: a comprehensive survey of retrotransposonsrevealed by the complete Saccharomyces cerevisiae genome sequence.Genome Res 1998, 8:464–478.

43. Pérez-Ortín JE, Querol A, Puig S, Barrio E: Molecular characterization of achromosomal rearrangement involved in the adaptive evolution of yeaststrains. Genome Res 2002, 12:1533–1539.

44. Hittinger CT, Gonçalves P, Sampaio JP, Dover J, Johnston M, Rokas A:Remarkably ancient balanced polymorphisms in a multi-locus genenetwork. Nature 2010, 464:54–58.

doi:10.1186/1471-2164-13-407Cite this article as: Peris et al.: Comparative genomics amongSaccharomyces cerevisiae × Saccharomyces kudriavzevii natural hybridstrains isolated from wine and beer reveals different origins. BMCGenomics 2012 13:407.

Submit your next manuscript to BioMed Centraland take full advantage of:

• Convenient online submission

• Thorough peer review

• No space constraints or color figure charges

• Immediate publication on acceptance

• Inclusion in PubMed, CAS, Scopus and Google Scholar

• Research which is freely available for redistribution

Submit your manuscript at www.biomedcentral.com/submit